Official commercial channel
Grok 4.5 API
xAI’s "Opus-class but faster and cheaper" pitch — and the list price backs it up.
Context
500K
Max output
64K
Serving precision
Vendor-managed
Licence
Proprietary
Grok 4.5 shipped to developers on 8 July 2026 with a 500K-token context window, text and image input, tool calling, structured outputs, web search and extended reasoning. Artificial Analysis ranks it #4 of 168 models on its Intelligence Index at a score of 54, while vendor list pricing sits at $2 per million input and $6 per million output ($0.50 for cached input) — noticeably below same-tier proprietary rivals. Worth knowing: the window is actually smaller than Grok 4.3’s 1M.
Proprietary, provided through the vendor’s official commercial licensing. Contract, invoice, audit trail — not an account pool and not a reverse-engineered client protocol.
Specifications
- Model ID
- grok-4.5
- Published by
- xAI
- How it is served
- Official commercial channel
- Licence
- Proprietary
- Context
- 500K
- Max output
- 64K
- Serving precision
- Vendor-managed
- Input modalities
- Text · Image
- Released
- 2026-07-08
- Price
- See pricing
What it is good at
- Near-frontier at a lower tier of price
- #4 on the Intelligence Index while listing at $2 input and $6 output per million tokens. For pipelines chasing "strong enough and economical", it is hard to leave out of the comparison.
- Cached input at a quarter of the rate
- Cache hits bill at $0.50 per million input tokens. Agent loops with long system prompts and heavy shared prefixes land well under the headline rate in practice.
- Web search is built in
- Extended reasoning and web search are native capabilities, so time-sensitive questions do not require you to stand up a separate retrieval pipeline first.
Where it fits
- Cost-sensitive pipelines that still want near-frontier capability
- Agents with long system prompts that hit cache heavily
- Q&A and research that depends on current information
- Cost/quality comparisons against the Claude and GPT flagships
Three steps
Change one line of base_url, put the model ID below in the `model` field, and leave the rest of your code alone.
model: "grok-4.5"
curl https://api.nexaix.net/v1/chat/completions \
-H "Authorization: Bearer $NEXAIX_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "grok-4.5",
"stream": true,
"messages": [
{ "role": "user", "content": "Summarise this contract." }
]
}'import os
from openai import OpenAI
client = OpenAI(
# the only new line
base_url="https://api.nexaix.net/v1",
api_key=os.environ["NEXAIX_API_KEY"],
)
# everything below is your existing code
stream = client.chat.completions.create(
model="grok-4.5",
stream=True,
messages=[
{"role": "user", "content": "Summarise this contract."}
],
)
for chunk in stream:
print(chunk.choices[0].delta.content or "", end="")import OpenAI from "openai";
const client = new OpenAI({
baseURL: "https://api.nexaix.net/v1", // the only new line
apiKey: process.env.NEXAIX_API_KEY,
});
// everything below is your existing code
const stream = await client.chat.completions.create({
model: "grok-4.5",
stream: true,
messages: [
{ role: "user", content: "Summarise this contract." }
],
});
for await (const chunk of stream) {
process.stdout.write(chunk.choices[0]?.delta?.content ?? "");
}About this model
Why is 4.5’s context smaller than 4.3’s?
It is: 4.5 ships 500K where 4.3 had 1M. xAI traded window for speed and token efficiency this generation. When you need a million-token window we also carry GLM-5.2, DeepSeek V4, Kimi K3, Claude Opus 5 and the GPT-5.6 family.
Do requests over 200K tokens cost more?
Yes. The vendor bills requests exceeding 200K total tokens at a higher tier. Build that step into your long-context cost estimates.
Is your Grok an official channel?
Yes. Like every proprietary model we carry, it comes through the vendor’s official commercial licensing — contract, invoice, audit trail — with no account pools and no reverse-engineered client protocols.
Also worth a look
- Official commercial channelClaude Opus 5A million tokens that never degrade, five-level effort control, zero data retention.claude-opus-5
- Official commercial channelClaude Fable 5Anthropic’s tier above Opus — the one you pick when the ceiling is the point.claude-fable-5
- Official commercial channelGPT-5.6 SolThe frontier tier of the GPT-5.6 family: 1.05M context, built for long-horizon agents.gpt-5.6-sol
Run your own eval on it
Take the trial credit, run the evals you already have against this model, and then decide whether it gets your production traffic.
